Oil Prices, Economic Growth and World Oil Demand

Andrew McKillop*

Introduction

Conventional economic reasoning, embodied in the notion of 'price elasticity' of demand, is that large oil price rises will necessarily cut oil demand and economic growth, perhaps resulting in zero economic growth, or recession. The higher the rise of oil prices, the faster it is assumed that 'price elastic' responses will play. Higher oil prices will "of course" reduce economic growth, may generate stock exchange panics and will produce inflation, leading to monetary and financial instability. Higher interest rates, and even a plunge into recession will then be needed to combat this through rapidly 'decompressing' the economy, thus cutting world oil demand, and reducing its price in a context where oil supply is always thought of as growing as fast, or faster than demand.

No received wisdom has any utility unless it can be demonstrated and proved, if it is have any more than the status of belief. The relation between oil prices and economic growth is in fact complex. The notion that price elastic responses describe inevitable, real and worldwide responses, and that any oil price rise necessarily and immediately reduces economic growth is in fact a travesty of the facts. From today's price levels for oil (around $30/barrel in the USA for light crudes), 'extreme' price levels would be needed before world economic growth fell. Until very elevated oil prices are achieved - probably well above $70/barrel in 2003 dollars - world economic adjustment mechanisms will always result in higher oil demand through the economic expansion, with or without inflation, that higher oil prices bring about at the world, or 'composite global' economic level.

Only extreme interest rates, as in the early 1980s in many OECD countries, can induce sufficiently violent and rapid contraction of economic growth to temporarily reduce oil consumption. With the return of growth, and even the near recessionary growth rates of many OECD economies, today, oil demand growth is always positive. So-called 'delinking' of oil from economic growth, in the early 1980s, was simply the interaction of intense economic recession and one-time energy savings due to energy conservations programs and polices engaged in the 1970s, after the Oil Shocks of that time.

Regarding the analysis and forecasting of world oil demand, the notion that 'classical' price elastic functions will apply when oil prices rise, even to levels around $50-$60/barrel (less than two-thirds the inflation adjusted price of 1979-1981, some 22 years ago), and using these to project future demand will necessarily result in incorrect demand forecasts. As already stated, higher oil prices, up to at least the $50-$60/barrel range, will almost certainly increase oil demand growth rates through the macoreconomic impacts of higher oil and energy prices at the world or composite global level.

Ineffective modelling of the real energy-economic adjustment mechanisms that come into play at times of fast rising oil prices, and after transition to higher general levels of oil and energy prices, can only give incorrect results. Bearing in mind that we are rapidly approaching Peak Oil (see below) the price factor in world oil demand trends should receive much greater attention.

Price elasticity, 'retrenchment' and adjustment

Price elasticity of demand is always thought of as negative - that is demand falls by a certain percentage for a certain percentage price rise of some factor input, in this case oil. Comfort for the belief that this 'function' should apply anywhere, in any economy, at any stage of economic development is obtained through narrowly focusing energy-economic performance of large OECD economies during times of fast rising oil prices.

Thus, for the period 1979-83, in which oil prices in nominal terms increased by about 115% (in the two years 1979-81), several large OECD economies showed year-on-year falls in oil consumption, by volume, that attained nearly 10% in certain quarterly periods, and well above 5% on an annual base. For example, in 1981 and 1982 volume oil demand of the US and Japanese economies respectively fell by 6.5% and 4.9% (USA) and 6.1% and 5.4% (Japan), all data being on a year-to-year base (Reference/ McKillop 1 and BP Stat Rev). However, it should immediately be noted that preceding the 1979-81 Oil Shock both economies showed consistent annual increases in their oil consumption (more than 3% for the US in 1978), in their 'adjustment and retrenchment' following the 1973-74 Oil Shock, in which oil prices had risen by about 295% to a level of about $42/barrel in dollars of 2003. Through the period of 1975-79, following a short and sharp initial downturn in economic growth and oil consumption growth rates, the USA, Japan, Germany and all other large OECD economies showed fast adjustment of their economies. Typical economic growth rates (real GDP, annual base) were in the 3%-4% region at that time, well over twice the typical growth rates of these countries today. By direct consequence - oil being the 'swing fuel' par excellence - their annual oil consumption growth rates were close, in percentage terms, to their percentage annual economic growth rates measured by increase of real GDP. This relationship, which can be called the 'oil coefficient of economic growth', was typically close to 0.75 (a 1% growth in real GDP entraining a 0.75% increase in oil demand by volume). An economy in which annual GDP growth rates and oil demand growth rates are closely similar can be called 'close coupled'.

It can be noted, here, that because of economic cyclic factors, and energy-economy 'structural' factors (including electrification and fuel mix changes), even OECD service dominated economies can become 'overcoupled', that is show a higher annual oil demand growth than their growth rates of real GDP, or very closely similar ratios, around 0.95. This is the current situation (since 1999-2000), in which the fall in real growth rates of several OECD economies has not at all entrained a fall to zero growth, or contraction of their oil demand. A case in point concerns the US energy economy in 2002-03. Despite very sluggish, and erratic quarterly economic growth trends (perhaps a composite annualized 1.75%-2% growth rate in real GDP terms), oil demand growth of the US economy was at an annualized rate of 2.9% in volume terms in the first 5 months of 2003. The US oil coefficient of economic growth, at this moment in time, is likely above 1 or 'overcoupled'.

Sequencing and impacts in the 'classic' Oil Shocks

In the period following the 1973-74 Oil Shock, during which nominal price rises for oil were about 295%, there was an initial 'price elastic' response, that is fall in oil demand growth rates, for nearly all OECD economies. In no case, however, did this stagnation or slight fall in oil demand exceed a 4% reduction in oil demand on an annual base. The period of 'decoupled' relationships between economic performance and oil demand did not exceed one year or four successive Quarters in any large OECD economy. As economic growth, oil consumption 'bounced back' rapidly, with typical oil demand growth rates approaching three-fifths to three-quarters the annual rate of real GDP growth.

This pattern was very different through 1979-83, following the second Oil Shock. In a context of extreme interest rates, the 'decoupled' period in which annual oil consumption fell, always with stagnant or falling real GDP, was very much longer for nearly all large OECD economies. The 'decoupled' period, in some cases, extended to about 36 months or 12 successive Quarters, that is from early 1980 to late 1983. Oil demand falls, for the OECD group, attained a composite 9% over the three years 1980-83.

Other factors contributed to this context of sluggish and hesitant economic adjustment, notably the accelerated electrification of many OECD countries, itself a policy response to the first Oil Shock of 1973-74 (Reference/ McKillop 2). However, after each of the 'classic' Oil Shocks of the 1973-81 period there was 'retrenchment' and then recovery. With the return to economic growth there was a return to annual rises in oil consumption within a certain period; the 'decoupled' period was much longer when oil prices had attained a level of about $103/barrel in 2003 dollars (in 1979), than in the first Oil Shock adjustment period when they had attained about $56/barrel in 2003 dollars (in 1974).

No extended 'decoupling' occurred after either Oil Shock. As might be expected in multiple series of world wide economic adjustments to factor cost and output changes, many 'thresholds' and 'triggers' were (and always are) in play. The fact that adjustment was much slower following the second Oil Shock can be considered normal given the oil price level achieved, and the extreme interest rates of the period, but the fact there was adjustment at all indicates the intrinsic and total dependence of 'classic' or conventional economic growth on increased oil consumption. Probably due to economic cyclic factors, and despite the extreme interest rates applied at the time in most OECD countries (a recessionary or defensive response aimed at cutting oil demand through triggering economic slump), final recovery in economic growth rates was very vigorous.

In 1984, with year-average oil prices in 2003 dollars still at about $57-$65/barrel, a level that would be considered 'extreme' today, the USA attained its highest ever annual economic growth (+7.5%, real GDP) in the entire 1945-2002 period. Many other OECD countries achieved relatively high levels of economic growth in a period extending to 1985-86. However, in the period of very low oil prices through 1986-1991 that terminated with Gulf War-1, economic growth rates fell very sharply throughout the OECD, with typically very slow growth of oil demand itself reinforcing the fall in oil prices. Analysed on a Quarterly base, for changes in oil price against changes in economic growth rates, the OECD group of countries, from the early 1990s to 1998-99, showed close coupling in the sense that each fall in economic growth rates was usually preceded by a fall in oil prices, and each small increase in average economic growth rates was usually preceded by a rise in oil prices.

The myth of 'decoupling'

This merely underlines that whatever the policy response to much higher oil prices (either recessionary and defensive, or expansionary that is 'adjustment through growth') the advanced industrial nations soon renew their dependence on oil consumption whenever their economies again start to grow. Without complete restructuring of the economy, food production, and transport systems, and de-urbanisation of population the world's economies will in fact remain 'coupled' with oil demand whatever the price, because of the complete dependence of modern urban-industrial economies on oil and oil products. That is, in other words, energy conservation or the reduction of energy demand through transition to a low energy economy will effectively be the only way that advanced urban and service economies of the OECD 'break the oil habit'.

Price elasticity and 'reverse' elasticity

Certainly in the last 50 years, economic growth as we know it has always required an increase in commercial energy consumption, and particularly oil and gas. The rate of economic growth is the prime determinant of the 'coupling' or relationship between the rate of economic growth, and rate of increase in oil and energy demand. Thus the interplay of generally declining economic growth rates in the OECD bloc and changes of economic structure (increasing services, electrification of the energy economy and ageing of the population) have all tended to reduce 'coupling' of OECD country economic growth, such as it is, with annual oil demand changes. This has, however, declined or even ceased since the middle-1990s. Several EU-15 economies, in the last 5-7 years, have shown 'overcoupling', that is higher oil demand growth rates than real GDP economic growth rates.

Comparing typical economic growth rates for OECD countries (here the G-7 group), and both 'traditional' and 'emerging' New Industrial Counties, the dramatic differences of economic growth rates are easy to see (Table 1), and these differences themselves explain much of the 'coupling' of oil demand with economic growth found in the NICs, and the decreasing 'coupling' in OECD countries.

TABLE 1 Economic growth rates (real GDP, annual average) for selected countries

 

Country	1968-79	1979-89	1989-95
South Korea	5.8%	3.7%	1.8%
Taiwan	10.0%	7.6%	6.3%
PR China	6.8%	9.8%	10.1%
India	3.5%	5.9%	4.4%
G-7 group	3.2%	1.5%	1.4%

It can be noted in passing that in the last period (1989-95) oil prices, except for a 4-month 'spike' preceding Gulf War-1 in 1991, sometimes 'downspiked' to inflation corrected price levels close to those of 1973. No increase or surge in economic growth was experienced in any OECD country. This tends to easily confirm the argument that cheap oil in no way facilitates or increases economic growth for the OECD countries !

As already noted, the notion of price elastic demand responses to much higher oil prices is based on classical economic thinking, and draws on data for rather short and selected periods of time, from a few large OECD economies for its support and demonstration. Absent from the underlying conception are the questions of actual economic growth rates, economic stage, ongoing industrialisation and urbanisation, the development of commercial energy supply and utilisation infrastructures and the economic demand-driven, increasing overall utility of oil within a growing economy. However, these elements are all critical for understanding 'reverse' or 'positive' elasticity of demand - that is increasing oil demand with increasing price. Classical price elastic notions are completely inapplicable for describing oil demand changes under regimes of much higher prices in fast growing New Industrial Countries as shown in Table 2, below

Table 2 Asian Tiger economic-driven, close-coupled adjustment to Oil Shock

Source/ BP Statistical Review, various edns

In the period covered by data in the above Table 2, nominal oil prices had increased by about 405%. While these New Industrial Countries (NICs) had very briefly reduced their oil demand through 1973-74, and again in 1979-80, overall oil consumption growth through the period of these 405% nominal oil price rises was very high. We can particularly note that through 1974-76, that is including the period of rapid 'bounce back' or recovery in economic growth and oil demand of the NICs, their growth rates of oil consumption, by volume, were in the 12%-15% range despite much higher oil prices continuing to 'work through' the pricing structure of economic inputs. The explanation, of course, is the economic stage of these economies (fast industrialising, export-oriented), and their economy-wide utilisation of oil and LPG-based energy using productive equipment, rather than electrified energy economies, typical of 'mature' OECD economies.

Economic stage

Concerning the energy economy, and avoiding any discussion of classical notions of 'economic stage', the most essential elements for modelling and forecasting oil demand focus the evolving shares of electricity, and oil, gas or coal based energy supplies and systems within an economy. This brings in the concept of Total Primary Energy, and Total Final Consumption (TFC), and also relates to questions of energy economic infrastructures. Briefly, we can characterise and contrast the NICs, and large, older or 'mature' OECD economies with very high levels of service sector output as a percentage of total GDP, by the second group of energy economies having a much higher part of electricity in their TFC (Reference/ McKillop 3). Construction and constitution of electricity supply systems, we can also note, requires very high levels of economic and energy investment. The claimed advantage of electricity-intensive economies, using nuclear, coal or other non oil primary sources for electricity production - that they 'cannot be held to ransom by high oil prices' - does not apply to their formative stages, and also sets rather high 'floors' to the 'compressibility' of energy demand of all kinds, including oil, at times of economic crisis (Reference/ McKillop 4).

On a world scale, and overall, there is about a 50/50 split in economic output between OECD countries, and all other nonOECD countries, including the new and giant NICs China and India. It is interesting to note that in terms of world oil demand, the split is about 55/45 in favour of (higher demand from) the OECD countries, immediately underlining the relative oil inefficiency of total economic output of the OECD economies. Even today, in early 2003 and after about 6 months of generally higher oil prices, economic growth rates are tending to increase in India and China, with inevitable oil demand increase by their close coupled economies. Schematically, and by reference to economic stage, we can present development stages, and oil demand driving factors of these stages, as summarised below in Figure 1

Figure 1 Energy economic stage oil demand characteristics

Drive elements in oil demand coupling

Again schematically, we can present the case of typical, 'classic economic development' for countries such as South Korea, India, Brazil or others, depending on the time frame used. At the earliest stage of development, sometimes extending back to the 1930s or before, their energy economic characteristics will include a high level of renewable energy sources utilised to support an agrarian, low income society, with very little urbanisation or industrialisation. Classical economic development may occur relatively fast, more slowly, or not at all (as is the case for many African economies). Only those countries which develop industrially, and urbanize are considered a 'success', and this model of development remains the basis of almost all loan financing and project formulation by agencies and institutions such as UNDP, ILO, the World Bank, and the regional banks.

In the case of South Korea we find that per capita oil demand, now approaching EU-15 per capita rates, has increased about 30-fold in 40 years. Without question, therefore, the oil demand of such an economy is close coupled in its early periods of fast growth and fast urbanisation and industrialisation. Increasing electrification, whether as a conscious 'oil dependence limiting' policy or not, will certainly help to reduce oil coefficients over time. However, in that part and period of the development process (in the case of South Korea from about 1965-85) where additional energy demand arises from the creation, and extension of energy dense infrastructures, oil coefficients of economic will remain close to unity, and in some cases can exceed unity. This can be understood from the example of the development of car manufacture and highway construction, as well as construction of nuclear power plants, in South Korea and other NICs. The mainly oil-based embodied energy of car manufacturing plant, highway construction equipment and materials, nuclear reactors, etc, both increases energy intensity of the economy while enabling (and forcing) economic agents and consumers to use more oil. Widespread 'motorisation' or the use of personal vehicles, in OECD and other countries, is associated with increases in personal or per capita oil demand of 10 - 20 times comparing on a 'before-and-after' basis.

On a world wide base the growth of the car industry and the world car fleet is a key vector for increasing energy-, and particularly oil-intensity of economic activity. The world's car fleet now stands at about 775 million units and is increasing at a world-wide average of about 6%/year, with the most rapid growth taking place in the new NICs, notably China and India, where double-digit annual growth rates are the norm. Through 1990-2002 the Chinese car fleet has increased at an average of more than 13% per year, growth of the car fleet attained more than 50% in 2001-2002, and national production will soon attain 6 Million units/year (Ref/ McKillop 5).

Given that classical economic development has always featured declining relative prices for and values of raw materials, unfinished products and energy, increasing oil consumption is virtually axiomatic for the development process of industrialisation and urbanisation that currently concerns at least one-half of the world's population. The time taken to transit from 'early' or oil-based and oil-intensive economic development, to 'mature stage', electrified energy-economic structures underlying the service-oriented economies of the ageing OECD countries is typically more than 20 years. Due to this, any forecasting of world oil demand trends without integration of these considerations will inevitably be incorrect and of little practical use.

Emerging oil price trends and future world demand

As noted by Campbell (Reference/ Campbell) we are moving quite rapidly towards Peak Oil, and may in the next 7 or 8 years attain the absolute limit of world oil production at around 83 Mbd. Even at the low aggregate rate of world oil demand growth that held in the 1990-2000 period, impressive total growth of oil demand was generated. In 1991, at the time of Desert Storm and the restoration of Kuwait to full sovereignty and full oil output, world demand was about 66.5 Mbd. Today it is about 79.5 Mbd, an increase of 13 Mbd, which is vastly more than Saudi Arabia will ever be able to supply however much it might want to deplete its oil reserves in the shortest possible time.

All downside action in oil pricing since March-April 2003, when prices shaved the $40/bbl 'psychological barrier', can be traced to over optimistic interpretations of Iraqi production and export potentials in the short- and longer-term, also bolstered by incorrect and unfounded analysis of world oil demand trends going forward. Recession trends notably in the US, German and Japanese economies were overvalued, while the pro-growth impacts of generally higher oil prices as a factor in economic growth were underplayed or ignored. Oil prices were marked down to the $25-per-barrel range following Baghdad's capture but have significantly rebounded since.

Today, the extent of pillage, arson and vandalism of key infrastructures either directly or indirectly needed to produce and export oil is wider known, and increasing every day. Forecasts made by Dick Cheney that Iraq may produce up to 2 Mbd by end-December 2003, allowing up to 1.45 Mbd of net exports after domestic needs, are very unlikely to be the real outturn. While prewar export numbers are highly variable and open to different interpretations, it is likely they attained over 2.25 Mbd (not including 'grey' and illegal exports, themselves around 0.4 Mbd).

Year-end net exports by Iraq will likely struggle to exceed 1 - 1.25 Mbd, and could in fact be much less, if gathering resistance to military occupation continues, and social chaos does not abate. As oil production difficulties for New Iraq become better known it is now clear that world supply has lost at least 1.5 Mbd at a time when demand is increasing by about 1.6 Mbd per year. At the same time, it is unsure that OPEC supply can easily be brought to a full 27.4 Mbd, with recent supply limitation accords, and difficulties for Venezuela and Nigeria all contributing to perspectives of OPEC supply being set in the 26 - 26.5 Mbd range, and increasingly difficult to raise above that level.

Firm demand and faltering supplies

On a year-average basis and for 2003, prices may well remain within the $25-$30/barrel range and then increase outside this range by or before winter 2003. If oil prices firstly remain 'firm', and then increase, perhaps to above $45/barrel, this will almost certainly be called an 'Oil Shock' by finance columnists and business observers. In the case of 'high' oil prices remaining a part of economic reality we can expect the energy economic mechanisms and factors, discussed above, to play a part in deciding and shaping future world oil demand trends, that is - on balance - underpinning demand.

The major reason for this is that oil demand by fast growing manufacturing and export activities in the NICs will tend to more than compensate any fall in oil demand by the OECD economies due to recessionary trends, partly (and to a very small degree) intensified by higher oil prices. For the oil demand driving factors of the NICs, the most important factor is solvent demand by low and middle income economies, for manufactured goods of all kinds. This demand is made solvent, and increased rather rapidly, by large and rapid oil price rises, through the impact of higher oil prices on world price levels for energy-intensive metals, minerals and agrocommodities. Prices for these commodities increase in line with oil prices, transferring wealth from OECD consumer countries to low and middle income country exporters. These latter countries, having increased revenue flows, address their newly solvent demand to the NICs, which now include one-third of the world's entire population in the shape of China and India. In addition, these latter countries have now entered the 'dynamic' of conventional economic 'takeoff' into energy intensive, urban-industrial economies, thus reinforcing their own domestic demand for oil in particular, as well as all other forms of commercial energy.

Leading indicators for this process include commodity price trends on a quarterly basis, volume movements of such raw materials, and orders received by NICs for their export goods. Using such indicators, and comparing these with 'ideal case' U-shaped demand-price oil coefficient curves (negative for OECD economies, 'reverse elasticity' for NICs), we can make good quality and above all realistic models for emerging world oil demand trends. For the OECD economies (see McKillop/2) we find that only oil price levels in 2003 dollars of above $60-$80/barrel induce either falls, or stagnation in oil demand trends over periods of about 36-40 months, before 'recoupling' occurs. For the NICs, conversely, much higher oil prices lead to an increase in consumption, because of the macroeconomic mechanisms sketched out, above.

Conclusions

While it may be conceptually pleasing, and undoubtedly saves time and work, to imagine that all energy economies subject to large and rapid oil price rises will exhibit 'classical' price elasticities of oil demand, simple facts and figures, and a few analyses, soon show that comparing the ageing, service oriented, electrified economies of the OECD countries with those of the NICs is like comparing chalk and cheese. Very different energy economic structures and mechanism are in play, producing entirely different oil demand outcomes under Oil Shock conditions. Given that world economic output is now close to 50/50 OECD/nonOECD it can be confidently said that 'classical' price elasticity functions used - often in elaborate models - by many official agencies and institutions are at least one-half wrong. Other methods, notably tracing raw material prices and industrial output by NICs are needed to build modelling procedures able to give reliable and realistic results.

The observed and simple fact that economic growth is strong, or even accelerating in the giant, emerging NICs of China and India, and is buoyant in several other large-population industrialising countries (eg. Turkey, Brazil, Iran), will ensure that oil demand growth rates at the composite or world level will be rather unlikely to fall below the so-called 'long term trend rate' of 1.8%/year, that held for the 1990-2000 period. Depending on oil price rises, and certainly with price levels of up to $50 - $60/barrel, composite or aggregate world economic growth rates are more likely to grow, than to shrink. In such case, overall world oil demand growth rates may well break out of the 'long term trend rate', that is used by OECD agencies such as the IEA to forecast future demand, and attain levels of considerably above 2%/year. World oil demand growth may well attain 2%-2.25% per year with prices at $50-$60/barrel.

* A McKillop

Former Expert, Policy and Programming, Divn A - Policy, DG XVII-Energy, European Commission

Founder Member, Asian Chapter, Intl Assocn of Energy Economists

REFERENCES

BP Stat rev: BP Statistical Review and BP Amoco Statistical review, various editions

McKillop 1: 'On Decoupling', A McKillop, Intl Jrnl of Energy Research, Vol 14, N° 1, 1990, J Wiley

McKillop 2: 'Improving the Quality of Oil Demand Forecasts', A McKillop, Oil & Arab Cooperation Quarterly, Vol 16, Issue 59, 1990, OAPEC, Kuwait

McKillop 3: As McKillop 1, above

McKillop 4: 'Energy and Economic development in Developing Countries', A McKillop, Oil & Arab Cooperation Quarterly, Vol 12, Issue 1, 1986, OAPEC, Kuwait

McKillop 5:'The Chinese Car Bomb' A McKillop (in) The Final Energy Crisis, Pluto Press, fothcoming

Campbell: 'The Assessment and Importance of Oil Depletion', C J Campbell, (in) The Final Energy Crisis, Pluto Press, forthcoming